pegasus-cnn_dailymail-finetuned-roundup

This model is a fine-tuned version of google/pegasus-cnn_dailymail on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.9488
• Rouge1: 55.4754
• Rouge2: 39.1074
• Rougel: 40.822
• Rougelsum: 47.3045
• Gen Len: 126.7778

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 16
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge1	Rouge2	Rougel	Rougelsum	Gen Len
1.8482	1.0	795	1.3177	51.1349	32.0791	34.4191	41.2001	127.0
1.362	2.0	1590	1.1975	52.1955	33.7858	36.2998	42.5368	127.0
1.2847	3.0	2385	1.1299	53.3694	35.9817	39.2437	45.2062	127.0
1.1673	4.0	3180	1.0903	53.3629	35.173	37.7775	44.3446	126.8889
1.0943	5.0	3975	1.0525	54.6538	37.3115	39.5883	46.8043	127.0
1.0615	6.0	4770	1.0317	54.6794	36.7147	39.5731	45.8527	127.0
0.993	7.0	5565	1.0144	55.2425	38.3984	41.1723	47.266	126.8704
0.9705	8.0	6360	0.9993	55.3351	38.3237	40.4765	47.3272	127.0
0.9266	9.0	7155	0.9864	55.174	38.0535	40.3484	46.7058	126.0926
0.9181	10.0	7950	0.9713	54.822	37.8024	40.0583	46.4962	126.5
0.9053	11.0	8745	0.9690	55.6235	38.9253	40.6261	47.1602	127.0
0.8513	12.0	9540	0.9614	55.3525	38.9343	40.4011	46.9631	127.0
0.8436	13.0	10335	0.9565	56.1316	39.5794	41.4653	47.7626	127.0
0.8343	14.0	11130	0.9522	55.7155	38.9484	40.968	47.0273	126.7778
0.8308	15.0	11925	0.9502	55.7195	39.1268	40.7967	47.1869	127.0
0.8296	16.0	12720	0.9488	55.4754	39.1074	40.822	47.3045	126.7778

Framework versions

• Transformers 4.23.1
• Pytorch 1.12.1+cu113
• Datasets 2.6.1
• Tokenizers 0.13.1